Internet and Gambling: Insights from Australia's NBN Rollout.

Gambling is a well-known leisure activity that leads to significant consequences when consumed excessively. We provide an analysis of the impact of access to faster and more reliable internet connection on gambling. We rely on variations in the rollout of Australia's largest infrastructure project, National Broadband Network (NBN) installation, to measure internet speed at the postcode level. Using gambling data from the Household, Income and Labour Dynamics in Australia (HILDA) survey, we find that access to high-speed internet is associated with a decline in gambling proxied by the Problem Gambling Severity Index (PGSI). However, a closer look at the various forms of gambling show that internet speed is associated with an increase in online-based gambling activities, which constitute a relatively small proportion of gambling activities that Australians participate in. In contrast, internet speed is associated with a decline in venue-based gambling activities, which constitute a large proportion of gambling activities that occur in Australia, and therefore explains the overall negative effect on gambling. We find that social capital and cognitive functioning are channels through which internet speed influences gambling.

Forecast evaluation for data scientists: common pitfalls and best practices.

Recent trends in the Machine Learning (ML) and in particular Deep Learning (DL) domains have demonstrated that with the availability of massive amounts of time series, ML and DL techniques are competitive in time series forecasting. Nevertheless, the different forms of non-stationarities associated with time series challenge the capabilities of data-driven ML models. Furthermore, due to the domain of forecasting being fostered mainly by statisticians and econometricians over the years, the concepts related to forecast evaluation are not the mainstream knowledge among ML researchers. We demonstrate in our work that as a consequence, ML researchers oftentimes adopt flawed evaluation practices which results in spurious conclusions suggesting methods that are not competitive in reality to be seemingly competitive. Therefore, in this work we provide a tutorial-like compilation of the details associated with forecast evaluation. This way, we intend to impart the information associated with forecast evaluation to fit the context of ML, as means of bridging the knowledge gap between traditional methods of forecasting and adopting current state-of-the-art ML techniques.We elaborate the details of the different problematic characteristics of time series such as non-normality and non-stationarities and how they are associated with common pitfalls in forecast evaluation. Best practices in forecast evaluation are outlined with respect to the different steps such as data partitioning, error calculation, statistical testing, and others. Further guidelines are also provided along selecting valid and suitable error measures depending on the specific characteristics of the dataset at hand.

Probabilistic Causal Effect Estimation With Global Neural Network Forecasting Models.

We introduce a novel method to estimate the causal effects of an intervention over multiple treated units by combining the techniques of probabilistic forecasting with global forecasting methods using deep learning (DL) models. Considering the counterfactual and synthetic approach for policy evaluation, we recast the causal effect estimation problem as a counterfactual prediction outcome of the treated units in the absence of the treatment. Nevertheless, in contrast to estimating only the counterfactual time series outcome, our work differs from conventional methods by proposing to estimate the counterfactual time series probability distribution based on the past preintervention set of treated and untreated time series. We rely on time series properties and forecasting methods, with shared parameters, applied to stacked univariate time series for causal identification. This article presents DeepProbCP, a framework for producing accurate quantile probabilistic forecasts for the counterfactual outcome, based on training a global autoregressive recurrent neural network model with conditional quantile functions on a large set of related time series. The output of the proposed method is the counterfactual outcome as the spline-based representation of the counterfactual distribution. We demonstrate how this probabilistic methodology added to the global DL technique to forecast the counterfactual trend and distribution outcomes overcomes many challenges faced by the baseline approaches to the policy evaluation problem. Oftentimes, some target interventions affect only the tails or the variance of the treated units' distribution rather than the mean or median, which is usual for skewed or heavy-tailed distributions. Under this scenario, the classical causal effect models based on counterfactual predictions are not capable of accurately capturing or even seeing policy effects. By means of empirical evaluations of synthetic and real-world datasets, we show that our framework delivers more accurate forecasts than the state-of-the-art models, depicting, in which quantiles, the intervention most affected the treated units, unlike the conventional counterfactual inference methods based on nonprobabilistic approaches.

The long-term course of alcoholism, 5, 10 and 16 years after treatment.

AIMS: To discover the long-term stability of drinking behaviour following an in-patient treatment episode. DESIGN: Three follow-up periods were used at 5, 10 and 16 years. The patients were classified as being abstinent, improved or unimproved on the basis of self-reported drinking behaviour. Patients who could not be interviewed at follow-up were classified as unimproved. SETTING: An alcohol dependence treatment programme at the University Hospital Tuebingen, Germany. PARTICIPANTS: We were able to locate all 96 patients at the 16-year follow-up. Seventy were alive and 26 had died. We collected information from 59 of the 70 surviving patients. The remaining 11 patients could be located and were definitely alive. FINDINGS: Thirty-eight of the 70 patients were abstinent, 10 were improved and 22 (including the 11 living patients without further information) were classified as unimproved. Our main finding indicates that the so-called 'improved drinking' is very inconsistent over time. In contrast, the abstinent and unimproved patients were much more stable in their drinking behaviour. CONCLUSIONS: This study extends our knowledge of the drinking trajectory and outcome from only a few years of follow-up to 16 years. Complete abstinence and unimproved drinking behaviour were the most stable drinking patterns observed over the long term, confirming study results obtained primarily from English-speaking countries.